Digital Communication Channel Equaliser using Single Generalised Neuron

Equalisation is necessary in a digital communication system to mitigate the effect of inter-symbol interference and other nonlinear distortions. A new reduced complexity approach to digital communication channel equalization is proposed based on a single generalised neuron (GN). Since it uses only a single GN, there is no problem of selection of initial architecture of the neural network giving optimum performance. It has less computational requirements giving rise to reduced training and computation time. The simulation results show that proposed equaliser bit error rate (BER) performance approaches to optimal Bayesian solution.